When a behavior is poorly represented in memory, respondents draw on estimation strategies to arrive at a frequency report. Given that memory declines with age, older respondents should generally be more likely to estimate than younger respondents. Counteracting this general tendency, some behaviors are more relevant for older than younger people and may therefore be better remembered. Experimental results support this reasoning. Compared to younger respondents, older respondents' frequency reports were more influenced by frequency scales when they pertained to mundane behaviors, but less influenced by frequency scales when they pertained to physical symptoms. Such differential effects of frequency scales invite misleading conclusions about age differences.